[Intel-gfx] [PATCH] drm/i915: Disable MI_STORE_DATA_IMM for i915g/i915gm

[Chris Wilson]

Quoting Ville Syrjälä (2017-09-06 16:00:49)
> On Wed, Sep 06, 2017 at 03:40:08PM +0100, Chris Wilson wrote:
> > The early gen3 machines (i915g/Grantsdale and i915gm/Alviso) share a lot
> > of characteristics in their MI/GTT blocks with gen2, and in particular
> > can only use physical addresses in MI_STORE_DATA_IMM. This makes it
> > incompatible with our usage, so include those two machines in the
> > blacklist to prevent usage.
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >  drivers/gpu/drm/i915/i915_drv.h            |  7 -------
> >  drivers/gpu/drm/i915/i915_gem_execbuffer.c |  7 ++++---
> >  drivers/gpu/drm/i915/intel_engine_cs.c     | 16 ++++++++++++++++
> >  drivers/gpu/drm/i915/intel_ringbuffer.h    | 12 +-----------
> >  4 files changed, 21 insertions(+), 21 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> > index 789f7502cd1f..6020a94daf81 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.h
> > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > @@ -4360,11 +4360,4 @@ int remap_io_mapping(struct vm_area_struct *vma,
> >                    unsigned long addr, unsigned long pfn, unsigned long size,
> >                    struct io_mapping *iomap);
> >  
> > -static inline bool
> > -intel_engine_can_store_dword(struct intel_engine_cs *engine)
> > -{
> > -     return __intel_engine_can_store_dword(INTEL_GEN(engine->i915),
> > -                                           engine->class);
> > -}
> > -
> >  #endif
> > diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> > index 53718a245ded..7687483ff218 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> > @@ -1168,6 +1168,9 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb,
> >               if (eb_use_cmdparser(eb))
> >                       return ERR_PTR(-EWOULDBLOCK);
> >  
> > +             if (!intel_engine_can_store_dword(eb->engine))
> > +                     return ERR_PTR(-ENODEV);
> > +
> >               err = __reloc_gpu_alloc(eb, vma, len);
> >               if (unlikely(err))
> >                       return ERR_PTR(err);
> > @@ -1192,9 +1195,7 @@ relocate_entry(struct i915_vma *vma,
> >  
> >       if (!eb->reloc_cache.vaddr &&
> >           (DBG_FORCE_RELOC == FORCE_GPU_RELOC ||
> > -          !reservation_object_test_signaled_rcu(vma->resv, true)) &&
> > -         __intel_engine_can_store_dword(eb->reloc_cache.gen,
> > -                                        eb->engine->class)) {
> > +          !reservation_object_test_signaled_rcu(vma->resv, true))) {
> >               const unsigned int gen = eb->reloc_cache.gen;
> >               unsigned int len;
> >               u32 *batch;
> > diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
> > index ae668340620e..ff56de4816cb 100644
> > --- a/drivers/gpu/drm/i915/intel_engine_cs.c
> > +++ b/drivers/gpu/drm/i915/intel_engine_cs.c
> > @@ -1378,6 +1378,22 @@ void intel_engines_mark_idle(struct drm_i915_private *i915)
> >       }
> >  }
> >  
> > +bool intel_engine_can_store_dword(struct intel_engine_cs *engine)
> > +{
> > +     int gen = INTEL_GEN(engine->i915);
> > +
> > +     if (gen <= 2)
> > +             return false; /* uses physical not virtual addresses */
> > +
> > +     if (gen == 3 && (IS_I915G(engine->i915) || IS_I915GM(engine->i915)))
> > +             return false; /* uses physical not virtual addresses */
> 
> nit: 'gen' check seems redundant.

I was optimistic that it would make gcc smarter. :)
-Chris